It has been suggested in opposition that temperate conditions may obtain even up to the edge of the ice, and that interbedded vegetal remains indicating temperate climate do not prove any considerable recession of the ice. The phenomena about existing glaciers have been appealed to in support of this demurrer. But the objection is not well taken. The climatic conditions which obtain about the borders of small, local glaciers, are not a safe guide as to climatic conditions which obtained about the margin of a continental ice-sheet, any more than the climatic conditions which obtain about a small inland lake are a safe criterion as to the climatic conditions about a sea-coast. The general principles of climatology, as well as specific facts concerning plant distribution, seem to us to indicate that the climate about the border of a continental ice-sheet must have been arctic. It is evident that the greater the distance north of the overlying drift remains of temperate plants are found, the more conclusive becomes the evidence. Plant remains indicating temperate climate at the very margin of the drift sheet which overlies them, would be less conclusive than similar evidences one hundred miles to the northward. It might be difficult to prove in any given instance that the ice which deposited the drift overlying plant remains advanced one hundred miles, or any other specific distance, south of any particular underlying forest bed. If the forest bed were continuous for the whole distance, the case would be clear. It would also be conclusive if the continuity of the drift overlying a forest bed at any point with that of a remote point to the south, could be demonstrated. In spite of these difficulties in its application, the vegetal beds constitute a valuable criterion in making the discriminations under consideration, when they are properly applied. Under proper circumstances the criterion may be conclusive when taken alone, and it may have corroborative significance when not itself conclusive. The absence of forest beds and of all traces of vegetal deposits whatsoever between beds of drift, is no proof of the absence of recurrent ice epochs, since the second advance of the ice might have destroyed all trace of the preëxistent soil and its vegetal life. It is always possible, too, that such beds exist, even if they have not been discovered. It would have been anticipated that they would not be abundant, or wide spread. The absence of forest beds is therefore at best no more than negative evidence. (2) _Remains of Land Animals._ Bones of mammalia or remains of other land animals, occurring in relations similar to those in which forest beds occur, may have a like significance. Their value as a criterion of separate glacial epochs is subject to essentially the same limitations as forest beds.